Approved some time ago

No, it is that the reference solution is consistently wrong when s == l[0] and len(l) > 1. For example:

l = [1, 2], s = 1

1 should equal 0

.

Fixed.

corrected now.

question,what if we have:

A) state, level, event = ., 5, O
 so we closed the gate, advanced to end (closed fully) and got strange "obstacle"
from security perspective I think it should not start immediately action to open

B) state, level, event = None, 0, O
 so we get obstacle before we could even think about moving returning etc.
for me that means we should not move and treat it as '.' - wait (stay on 0 level)

C) what if we had state machine with "error"
input = "OOPOO" 
output= "00100"
if we had obstacle and never get "freed" from it, I think the "safe" should be ".P" only starts the gate close progression
 so assume "0" * 5:
output= "00000"
would be correct then

it's kinda tricky, but want to know if you thought about consequences of this exercise "in the wild"

FYI @chainsaw-maniac , @brendanvos

Guess what, your JS solution ported to Python times out on the exact same test case.

@Akar-0, Is your Python solution based on your JS solution, then I'll try to translate that one and test ...

Translating the author's Ruby solution to Python gives me the exact same timeout as my own submission. This is too weird for words.

Reference solution passes in ≃ 5 seconds (apparently without special trick that would come from knowing in advance the tests design).

Can this still be solved in Python? I (should) have a very fast solution, but I'm no Python expert.

My old Python solution times out now. Maybe the current runner is slower. It's the problem with performance katas: they are closely dependent on language and test versions.